Resources
Resources
Collections
Compliance and QA
Banking and lending
AI

Your collection agency is working your best accounts. Who is working the rest?

Resources
Resources
Collections
Compliance and QA
Banking and lending
AI

Your collection agency is working your best accounts. Who is working the rest?

Collections
Compliance and QA
Banking and lending
AI

Your collection agency is working your best accounts. Who is working the rest?

In a typical primary placement with a third party debt collection agency, fewer than 40% of accounts receive anything resembling sustained collection activity. The rest age, and while they age, collectability erodes at roughly 1% per week.

The industry has a term for this, the contingency alignment problem. Agencies operating on a contingency model rationally concentrate effort on accounts most likely to produce recoveries.

Fresh placements, elevated balances, high-contact-probability consumers get the attention, while low-balance, aged, or difficult-to-locate accounts move to the back of the queue, then drop off entirely.

11–21%
Industry average debt collection liquidation rate
Source: Fair Capital / IC System industry surveys
~40%
Typical portfolio penetration at a third-party agency — most accounts never get meaningful contact
Source: NeuAnalytics vendor management data
~1%/week
Collectability decline rate for every week past the due date
Source: Fair Capital industry data

Why contingency models produce uneven portfolio coverage

The FTC's study of the debt buying industry documented the primary, secondary, and tertiary placement cascade.

Accounts go to a primary third party debt collection agency, the unrecovered balance moves to a secondary, and what remains passes to a tertiary tier at progressively higher debt collection agency fees and diminishing recovery probability.

Each handoff compounds the loss, and collectability declines sharply after the first 90 days. By 12 months past charge-off, recovery rates have fallen to roughly 25% of their initial level.

What the data also shows, and what vendor scorecards rarely capture, is that within any single placement, agencies apply their own internal prioritization. Accounts are not worked uniformly.

Contact cadences, collector assignment, and dialing intensity all follow the economics of contingency. Maximum effort goes where expected revenue per hour worked is highest. The accounts that receive that intensity are a fraction of what was placed.

How debt collection agency fees make the problem worse

Industry-standard debt collection agency fees run 15 to 25% on fresh placements under 90 days, climbing to 25 to 40% in the 6 to 12 month range and 40 to 50% on legal-track or significantly aged paper.

For a mixed-age portfolio, blended debt collection agency fees typically settle between 20 and 35%. What receives less scrutiny is their cumulative impact when measured not against gross recovered dollars, but against total face value, including the face value of accounts that never received meaningful activity.

Industry data
Debt collection agency fees by debt age
Typical low-to-high range of agency fees as debt ages — the older the account, the more you pay
Sources: Southwest Recovery Services · IC System · Fair Capital · Empire Debt Collection

Hypothetical portfolio math
$100M face value portfolio — one agency, one cycle
Accounts the agency actively works
35% of book
Liquidation rate on those active accounts
12%
Gross collected
$4.2M
Paid to agency at blended 28% contingency
$1.17M
Your net recovery
~$3.0M
Unworked face value sitting untouched
$65M

The agency sends you a check for $3M and a clean report. Meanwhile $65M in face value is losing ~1% collectability every week it sits untouched.


The agency remits a check, and the rate looks within range. But the effective cost of debt collection agency fees, measured against the full recovery potential of the book and not just the portion the agency pursued, tells a materially different story.

As debt ages within the placement cycle, that calculus worsens. Fee percentages climb on older accounts precisely because recovery probability is falling, while actual collection effort per account contracts. The buyer pays more for less, on a portfolio whose recoverable value is declining weekly.

The structural problem
As debt ages, debt collection agency fees rise — but effort drops
Agency fee % increases on older accounts while actual contact attempts per account decline
Fee % rises — agencies charge more on older debt because recovery is harder
Effort drops — agents spend fewer attempts per account as age increases, compounding the problem
Illustrative based on industry-reported contingency rate ranges and collection activity patterns

{{cta-banner}}

Running the full portfolio before third party debt collection agency

Running first-party collection at meaningful scale required licensed infrastructure and management overhead that the economics did not support outside of the largest buyers.

The choice was between delegating to agencies and absorbing debt collection agency fees and coverage gaps, or internalizing operations at costs that rarely penciled out. AI agents alter that calculus in a way that is less incremental than categorical.

An AI agent operating across a debt portfolio has no marginal cost per contact, no economic rationale to deprioritize sub-$500 balances or 18-month aged paper, and no reason to concentrate effort on a subset of what was placed.

It works the full book, every account on a cadence calibrated to its risk profile, before any placement decision is made. Every interaction is documented, compliance is verifiable on every call, not on a statistical sample.

Coverage model
Work the whole book first — then send the rest to agencies
Illustrative approach for a $100M face value portfolio
Current model — full placement with agency
📋
$100M placed with agency
Full book, day one
🏢
Agency works top 35%
Cherry-picks high-value accounts
65% barely touched
Collectability decaying weekly
~$3M net to you (after fees)
28% fee on every dollar
$65M unworked face value
AI-first model
New model — AI agents first, agencies for the remainder
🤖
AI agents cover 100% of book
Every account, consistent cadence, 0% contingency
🏢
Agencies get the hard remainder
Complex negotiations, legal escalation
More recovered — keep 100%
$0 fees on AI-collected dollars
Agencies used for what they're good at

This isn't about cutting off your agency network. It's about changing the order of operations. AI agents collect everything that doesn't require human negotiation. Agencies handle what's left — which is genuinely the work that benefits from human judgment.

What changes in practice

1. Compliance moves from probabilistic to verifiable

Every conversation is recorded, transcribed, and reviewed against FDCPA requirements and internal protocol. The posture shifts from reactive to proactive, with call-level visibility before any dispute materializes rather than surfacing problems through CFPB complaints or periodic audits after the fact.

2. The data exhaust stays with the buyer

Contact rates by segment, objection patterns, right-party contact performance, and payment intent signals accumulate as proprietary intelligence on how the portfolio behaves. Agencies provide reporting, while AI agents produce a continuous, compounding signal that informs collection strategy and future portfolio decisions.

3. Agency placements become more targeted

When accounts go to agencies, they arrive with a documented history of contact attempts, consumer engagement data, and pre-qualification for the type of work agencies perform well. The placement is more precise, the paper is better characterized, and debt collection agency fees apply only to the work that genuinely requires them.

Recovery comparison
Agency-only model vs. AI-first model
Illustrative $100M portfolio — what changes when AI agents run first

Illustrative. AI-first model assumes AI agents collect low-to-medium effort accounts; agencies handle complex remainder. Fees apply only to the agency-placed portion.

Questions to ask before your next agency review

Before the next agency performance conversation, pull these numbers. If most of them are unavailable, that is itself informative.

Self-Audit
Portfolio penetration diagnostic
Pull these numbers before your next agency performance review
1
What % of placed accounts get at least one live contact attempt in the first 60 days?
Dial attempts don't count. Live conversations. If you can't get this number easily, your agency isn't giving you the data you're owed.
2
How does your liquidation rate differ between the top and bottom 20% of balance bands?
A significant gap is evidence of skimming. If large-balance accounts liquidate at 2–3x the rate of small-balance accounts, you know exactly where the effort is going.
3
What is your effective fee tax on dollars collected?
Take total contingency fees paid over the last 12 months and divide by gross recoveries. If you're paying more than 30 cents on every dollar the agency collects, ask yourself what it would cost to deploy that capital differently.
4
How many accounts are being returned or re-placed with minimal documented activity?
Accounts returned after one placement cycle with no contact record are accounts the agency gave up on. How many of those are in your portfolio right now?
5
What visibility do you have into compliance on individual calls?
If your answer is "we review a sample," that's a compliance risk as much as an operational one. Full call-level visibility is standard now. You should be demanding it.
6
What happens to your performance data if you terminate the agency relationship?
If the answer is "we lose it," that's not a vendor relationship — that's a dependency. Your collection data should belong to you.

{{cta-banner}}

Collections
Compliance and QA
Banking and lending
AI
Cover full portfolio before placement
Every account worked, every call documented, before a single dollar goes to a contingency.
See proAgent